Articles / Aufsätze
The third gender of Old Italian*
Michele Loporcaroa, Vincenzo Faraonia and Francesco Gardanib
aUniversität
Zürich / bWirtschaftsuniversität Wien
We demonstrate that Old Italian had a three-gender system within which the
neuter still qualified as a fully fledged gender value. To substantiate this claim,
we adduce evidence showing that (a) Old Italian had three distinct sets of controllers, each of which selected a separate agreement pattern; (b) to each one of
those three controller sets, including the neuter, nouns were assigned belonging
to different productive inflectional classes and (c) the neuter still selected at least
one dedicated agreement formative, thereby still displaying traces of its original
status as a target gender. This novel evidence from Old Italian squares well with
what is known about past stages of other Romance varieties. Also, we briefly
address the consequences of our results both for a reconstruction of the LatinRomance transition and, more broadly, for the theoretical and methodological
approach to the study of the diachronic development of gender systems.
Keywords: target vs. controller gender, neuter, agreement, inflectional class,
productivity, Romance diachronic morphology
1. Introduction
Romance languages generally display a binary gender contrast (masculine vs. feminine), leading many scholars to project this situation back onto Proto-Romance,
thus assuming that a reduction of the original three-gender system of Latin had
already occurred by that stage via the demise of the neuter. We show that Old
Italian (i.e. Old Tuscan)1 had a three-gender system within which the neuter still
* This paper was conceived and written jointly, though for academic purposes §§5 and 6 must
be ascribed to ML, §§3 and 4.2 to VF and §§1, 2 and 4.1 to FG. We are indebted to Matthias
Grünert, Joe Salmons and three anonymous reviewers for comments and discussion. The usual
disclaimers apply.
1. The label ‘Old Italian’ refers, by a well-established convention reflected, for instance, in the title of the recent reference grammar by Salvi & Renzi (2010), to the Tuscan dialects — Florentine,
Diachronica 31:1 (2014), 1–22. doi 10.1075/dia.31.1.01gar
issn – / e-issn – © John Benjamins Publishing Company
2
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
qualified as a grammatical gender in its own right. To substantiate this claim, we
adduce evidence showing that a) there were three distinct sets of controllers that
each selected a separate agreement pattern; b) to each one of those three controller sets, including the neuter, nouns were assigned which belonged to different
productive inflectional classes (as defined in §4.1) and c) the neuter still selected
at least one dedicated agreement formative — i.e., a formative not syncretic with
either masculine or feminine inflections — thereby still displaying traces of its
status as a target gender as it was in Latin.
These properties of the gender system of Old Italian, we contend, have been
poorly understood so far, clouded by the widely held assumption that the change
from three to two genders occurred already in Proto-Romance / Late Latin. The
Old Italian facts are incompatible with this assumption and cumulate with evidence from other Romance branches to substantiate the reconstruction of an
intermediate stage between the fully fledged Latin three-gender system and the
modern Italian binary system.
The paper is organized as follows: in §2, we introduce some basics for the analysis of gender systems, illustrating them with examples from Romanian, which is
the most conservative among the modern standard Romance languages in this
respect and thus proves instrumental for the introduction of the diachronic issue
concerning the intermediate steps in the development from the Latin to the modern Romance gender system. In §3, we briefly address contemporary Italian, then
consider Old Italian in §4. This section is bipartite: §4.1 enumerates the inflectional classes associated with the neuter gender in Old Italian, showing that some
of them were productive throughout the Middle Ages, whereas §4.2 discusses the
remnants of dedicated neuter agreement formatives still observable in medieval
Florentine (or, in other words, residues of the third target gender; see §2) and compares them with their precursors in Late Latin texts from Tuscany. Finally, §5 deals
with the comparative Romance picture, briefly addressing varieties other than Old
Florentine, to show that the persistence of dedicated neuter agreement was more
robust and lasted longer in those systems. Our main goal is a new description of
in the first place — as documented in the late Middle Ages. While the earliest Florentine text
dates back to 1211 (see Castellani 1958: 19–95) and the earliest vulgar document from Tuscany
— the so-called Conto Navale Pisano (see Baldelli 1973: 5–33, Castellani 1976: 123–148) — to
the early 12th century, the rise of Florentine to the status of a shared standard language for the
entire Italian peninsula is a later phenomenon, which was accomplished between the 15th and
the early 16th centuries for formal usage, and only during the 20th century, after the political
unification of the country, for everyday spoken use. Thus, for the medieval period focused on
here, Old Florentine — also called, retrospectively, Old Italian since it is formally the ancestor
of the modern standard language — is a daughter language to Latin on a par with, say, Old
Neapolitan, Old Milanese, Old Venetian, etc.
The third gender of Old Italian
the gender system of Old Italian; however, we argue in the conclusion (§6) that our
investigation has broader methodological implications.
2. Analyzing gender
To pave the way for our discussion of Old Italian, we first have to introduce minimal analytical tools starting with the definition of gender:
(1) “Genders are classes of nouns reflected in the behavior of associated words”
(Hockett 1958: 231)
Another notion important for present purposes is Corbett’s distinction (1991: 151)
between controller and target gender:2
(2) “We should … differentiate controller genders, the genders into which the
nouns are divided, from target genders, the genders which are marked on
adjectives, verbs and so on.”
This distinction is illustrated by Romanian, the only modern Romance standard
language for which most current analyses, including Corbett’s (1991: 151), agree
on assuming a three-gender system. Consider the data in (3), which illustrate gender agreement on both definite articles and (first class) adjectives:
(3) singular
a. student-ul e bun
“the student is good”
b. vin-ul e bun
“the wine is good”
c. băutur-a e bun-ă
“the drink is good”
plural
studenţi-i sunt bun-i
“the students are good”
vinuri-le sunt bun-e
“the wines are good”
băuturi-le sunt bun-e
“the drinks are good”
masculine
neuter
feminine
Although articles and adjectives only display two distinct sets of forms, one for
the masculine (-ul/-i) vs. one for the feminine (-a/-le), three different agreement
patterns emerge. These identify three distinct sets of nouns satisfying the definition of gender in (1). The class exemplified in (3) by vin “wine” displays gender
agreement like the masculine nouns in the singular and like the feminine nouns
in the plural — a combination of agreement forms which differs from those used
for masculine and feminine. Thus, by the definitions (1)–(2), the Romanian neuter
qualifies as a third controller gender, also defined more traditionally as a genus
2. We use these terms, which have both forerunners (Hockett’s 1958: 230 selective vs. inflectional gender) and successors (Corbett’s 2011: 459–460 non-autonomous vs. autonomous gender), since they are better anchored in current studies in morphological theory.
3
4
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
alternans (or ‘alternating gender’), given the alternating syncretisms in singular
vs. plural highlighted in (3). Its status as a full-blown gender is confirmed by the
fact that the same gender agreement pattern exemplified by vinul is selected by a
large number of nouns which belong to several inflectional classes, some of which
are productive in contemporary Romanian. It can be added that, unlike in other
Indo-European languages such as German, the Romanian neuter is semantically
coherent as it includes exclusively nouns denoting inanimate entities (apart from
class-denoting terms such as animal).3
Given a three-gender analysis of Romanian, a further issue that arises is that
of the diachronic relationship between the Latin and Romanian gender systems.
While both languages have three genders, this, of course, does not imply that
nothing has changed, since only in Latin were there three distinct sets of gender
agreement targets, which reduced to two in Romanian. While Latin had three target genders, Romanian has (only) maintained three distinct controller genders.
This can be interpreted as an intermediate stage on the path to the binary gender contrasts otherwise observed in the modern Romance languages. Our analysis
of Old Italian — to which we turn now — lends comparative support to this diachronic reconstruction.
3. Italian
Let us move on to the Italian facts. Referring back to properties (a) to (c) listed
in §1, all were satisfied in Latin, including, crucially, (c), since neuter agreement
was signaled in several paradigm cells by dedicated inflections on determiners,
adjectives, participles, etc. (‘associated words’ by the definition in (1)). Instead,
in modern Italian, not only (c) but also (b) — viz. the existence of productive
noun inflectional classes associated with the neuter — have long been lost. As
for (a) — viz. the occurrence of three distinct sets of controllers selecting each a
separate agreement pattern — the picture is not so clear-cut. In fact, some scholars
(e.g., Merlo 1952; Bonfante 1961, 1964, 1977) have maintained that Italian too, like
Romanian, does preserve a neuter gender, based on the three agreement patterns
observed in (4):
3. A three-gender analysis of Romanian is maintained by, e.g., Graur (1928), Bonfante (1964,
1977), Jakobson (1971: 187–189), Aikhenvald (2000: 45–46), Matasović (2004: 51–52), Igartua
(2006: 60–61), Acquaviva (2008: 135–140). Two-gender analyses have also been proposed by,
e.g., Hall (1965: 425) and some others (cf. §6).
The third gender of Old Italian
(4) singular
a. il naso è lung-o
“the nose is long”
b. il braccio è lung-o
“the arm is long”
c. la gamba è lung-a
“the leg is long”
plural
i nasi sono lungh-i
“the noses are long”
le braccia sono lungh-e
“the arms are long”
le gambe sono lungh-e
“the legs are long”
masculine
?
feminine
As in (3), gender agreement is signaled by the determiner forms and the inflections
of the first class adjective lungo. Formally, (4) is identical to (3). In particular, nouns
like il braccio / le braccia (4b) behave like Romanian neuters such as vinul / vinurile
in (3b) above. In spite of this, there is broad consensus that the Italian type il braccio
/ le braccia does not constitute a third distinct gender of its own (see, e.g., Dressler
& Merlini Barbaresi 1994, Dressler & Thornton 1996: 5, D’Achille & Thornton
2003). The argument adduced is that in contemporary standard Italian, unlike in
Romanian, this kind of agreement is limited to one inflectional class with a small
number of members and has been progressively eroded over the last centuries.4
On the strength of this evidence, it is legitimate to regard modern Italian as
a two-gender system, as is standard in reference typological work (e.g., Corbett
2006: 245). Therefore, if contemporary Italian, unlike Latin, does not have a third
gender, a change from three to two genders must have occurred; the issue is when
and how this change came about. In historical linguistics, issues of this kind are
recurring and often difficult to solve due to the fact that, in general, the description
of diachronic change depends upon the analysis of the discrete synchronic states
corresponding to the input and the output of the change, and this analysis is likely
to pose problems. Thus, for instance, the ascertainment of the number of values of
a given feature may be no simple task, a point nicely made in Corbett’s (2008: 1)
discussion of case values in contemporary Russian. As for change in the number
of values of the feature gender, one possible scenario is that one of the genders,
before being eliminated outright, is depleted of lexemes to the extent that it can be
labeled an ‘inquorate gender’ (Corbett 1991: 170ff.), that is, a gender which lacks
a critical mass of lexemes to allow considering it a wholly legitimate value of the
4. We will not discuss further analyses of the data in (4b) in terms of derivation (Acquaviva
2008) or number (Togeby 1952). For discussion, see Loporcaro & Paciaroni (2011: 401–403). In
(4b), this agreement pattern is exemplified with just one inflectional class. Note, however, that in
contemporary Standard Italian, some nouns belonging to other inflectional classes (il carcere/le
carceri “jail”, il gregge/le greggi “flock”, as well as l’arancio/le arance “orange”, l’orecchio/le orecchie
“ear”) also select the same agreement pattern. While this shows that the alternating agreement
pattern (4b) still has some residual force of attraction, it hardly changes the overall picture since
none of those lexemes belongs to productive inflectional classes.
5
6
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
gender feature. This is indeed what Igartua (2006: 60) maintains for the modern
Italian type il braccio / le braccia.
In order to understand what occurred in the development from Latin to
Italian, we need to survey the development of the noun system and, in particular,
the synchronic situation in Old Italian, with which the next section is concerned.
4. The gender system of Old Italian
For the sake of perspicuity, we address separately first the Old Italian evidence for
(the productivity of) inflectional classes featuring nouns whose gender value was
the neuter (§4.1), and then the evidence for (residual) dedicated neuter gender
agreement (§4.2).
4.1 Productivity of the inflectional classes tied to the neuter gender
In Old Italian, there were several inflectional classes whose members selected alternating agreement of the kind seen in (4b).5 Consider the full array of noun
inflectional classes in Old Italian:6
(5) Noun inflectional classes in Old Italian and their relation to gender:
Class
-o / -i
-a / -i
-e / -i
-a / -e
-o / -a
-o / -ora
-e / -ora
-o / -e
invariable
Examples
lo libro / li libri
lo poeta / li poeti
lo fiore / li fiori
la siepe / le siepi
la casa / le case
lo dito / le dita
lo prato / le pratora
lo nome / le nomora
lo pomo / le pome
la unghia / le unghia
lo dì / li dì
Gender value
masculine
feminine
genus alternans
feminine
masculine
Gloss
“book”
“poet”
“flower”
“hedge”
“house”
“finger”
“lawn”
“noun/name”
“apple”
“nail”
“day”
5. This agreement pattern is exemplified in (5) in the greyed-out cells of the second column;
note the older form of the singular definite article lo dito instead of modern Italian il dito (lo is
nowadays selected only before initial geminate consonants and consonant clusters that are not
exhaustively syllabified as onsets: e.g. lo spreco “the waste”).
6. For different arrangements of the inflectional classes in Old Italian, see Penello et al. (2010: 1389–
1397), Gardani (2013: 422) and D’Achille & Thornton (2003: 212). The display in (5) is purely qualitative and abstracts away from differences in quantitative robustness between the paradigms listed.
The third gender of Old Italian
The crucial classes for our present discussion are greyed out in (5). The first, dito /
dita, has survived into the modern language though, as we said, it has been losing
members during the last centuries. However, it had earlier been productive for
several centuries. Gardani (2013: 407) shows that the classes dito / dita and prato
/ pratora still display a fair degree of productivity between 1100 and 1400, a claim
substantiated by the fact that the inflectional class dito / dita a) was assigned to
deverbal nouns derived via conversion, e.g. grido / grida “scream” (first attested
before 1292) from gridare, urlo / urla “scream” (14th century) from urlare and b)
triggered class shift of nouns such as anello / anella “ring” (before 1292) from earlier anello / anelli < Latin anellus -i and crino / crina “horsehair” (1282) from crine
/ crini < Latin crinis -is. This satisfies the definition of productivity adopted there,
as the force of attraction that inflectional patterns exert on new lexemes, both foreign and native, as well as on pre-existing paradigms of native lexemes (Dressler
2003: 31, Gardani 2013: 39).
If we now consider productivity as an indicator of vitality of inflectional classes, and the genders fed by productive classes as vital genders, then we must assume
for Old Italian three controller genders — viz. masculine, feminine and neuter
(the genus alternans associated with the greyed-out cells in (5)), as shown by the
following scheme, where gender agreement is exemplified with the (nominative,
for Latin) endings of 1st class adjectives:
(6) a. Latin
b.
Old Italian
sg
pl
sg
pl
m
-us
I
-i
m
I
-i
-o
n
-um
II
-a
>
III
-e
f
-a
III
-ae
f
-a
II
Two observations are in order here. Firstly, the evolution summarized in (6a–b)
(supported by the data in (5)), combined with the Romanian facts in (3), is at odds
with the widely held assumption that the original Latin gender system had already
shrunk to a binary contrast by Late Latin / Proto-Romance (e.g., Schön 1971: 4;
Tekavčić 1980: II, 66; Alkire & Rosen 2010: 192; Rovai 2012: 116). Secondly, reconstructing the change from (6a) to (6b) — i.e. from a three-target-gender system
to a three-controller-gender system — implies nothing as to exactly when that
change occurred. We shall now see in §4.2 evidence that this change in Old Italian
was completed much later than is usually assumed.
4.2 Remnants of neuter target agreement in Old Italian
A circumstance overlooked until Faraoni et al. (2013) is that in Old Tuscan texts
from the 13th–14th centuries, nouns of the third gender in (6b), which usually
7
8
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
selected a feminine agreement formative on plural targets (e.g. labbra vermiglie
“lips(n) red:f.pl”), could still trigger a dedicated agreement realized by a plural
formative -a on adjectives and determiners, as shown in the following examples:
(7) a. li denti minotetti / di perle son serrati; / lab[b]ra vermiglia, li color’ rosati
“the small teeth / of pearls are made; / lips vermilion, / the colors like
roses”
(Chiaro Davanzati, second half of the 13th century, Florentine;
Menichetti 1965: 137)
b. a guardare le detta castella et cassari
“to guard the said castles and turrets”
(Statuti Senesi 1309–1310 (Gangalandi); Lisini 1903: I, 219)
c. la grave e continua spesa che quella mura richeggiono
“the heavy and continuous costs that those walls require”
(Lettere volterrane, 1348–1353; Della Valle 1982: 201)
d. poi che furono entrate nella letta, ciascuna s’infinse di volersi levare a dire
certe orazioni
“after they had got into the beds, all of them pretended that they wanted
to stand up and say certain prayers”
(Matteo Corsini, 1373, Florentine; Polidori 1845: 104)
e. Fuggiamo quinci acciò che non ci rovinino la bagnora addosso, ne’ quali …
“Let us flee from here, lest the baths should fall upon our heads,
wherein(pl) …”
(Leggenda Aurea, second half of the 14th century, Florentine; Levasti
1924–1926: I, 119)
f. tutta la borgora di Melano misse al fuoco
“all boroughs of Milan did he set on fire”
(Leggenda Aurea, second half of the 14th century, Florentine; Levasti
1924–1926: III, 1580)
g. e scaricarono la sacca. Scaricate che l’ ebbono …
“and they deposited the sacks. Once they had deposited them …”
(Sacchetti, Trecentonovelle, second half of the 14th century, Florentine;
Pernicone 1946: 529)
h. acciò vadino nella castella e nelle ville che son dintorno
“so that they go to the castles and the lands around”
(Bibbia (09), 14th–15th century, Tuscan; Negroni 1886: 84)
The paucity of such attestations requires a word of caution. Given that this agreement pattern is realized by just one final vowel, it cannot be excluded a priori that
one or the other of the examples in (7) is illusory — i.e. has arisen as a copying error. However, the hypothesis of mere material errors becomes less plausible when
this kind of agreement appears on several different agreement targets (e.g., tutta
The third gender of Old Italian
la borgora in (7f)). On the other hand, plural agreement on verb forms ((7c, e))
and (relative) pronouns ((7e)) guarantees that what occurs there is the realization
of a genuine (neuter) plural rather than a collection of instances of the well documented reanalysis of neuter plural as feminine singular (e.g. Italian la foglia “the
leaf(f):sg”, from Latin illa folia “those / the leaves(n):pl”).
Further confirmation of the value of the Old Tuscan evidence just discussed
comes from considering the data provided by other branches of medieval ItaloRomance and, more generally, from the inspection of the overall Romance picture, to which we shall turn in §5. Before this, however, we shall take a step backwards and reconsider the Latin-Romance transition: obviously, the diachronic
source of the agreement formative -a in (7) is the Latin neuter plural ending -a
(e.g. bona “good:n.pl”), which apparently persisted into Italo-Romance. Indeed,
our Old Tuscan data provide evidence for Kuryłowicz’s (1964: 212) reconstruction.
Departing from the widely held idea that the change from a three- to a two-gender
system had already taken place in Proto-Romance, Kuryłowicz postulated for a
preliterary stage of Old Italian a gender agreement system along the following lines:
(8) Gender agreement (in 1st class adjectives and the definite article) in
preliterary Italian: (Kuryłowicz 1964: 212)
sg
pl
m
-o
-i
f
-a
-e
n
-o
-a
While such a stage, by definition, is not directly documented, some further hints
add to the residual evidence in (7) and confirm that (8) must have been the case.
Such hints can be gained by the inspection of Late Latin texts. For example, in
the texts collected in the Codice Diplomatico Longobardo (CDL)7 dating from the
7th–8th centuries, the type ill-a brachi-a “th(os)e-n.pl arm(n)-pl” is abundantly
documented. The following examples are just 3 of 68 occurrences found in 209
legal documents from Tuscany:8
(9) a
b
ista altaria
“these altars”
per loca designata
“through the said places”
(CDL I, 62, line 19)
(CDL I, 102, line 20)
7. On those texts and their value for the reconstruction of the prehistory of Italian, see in particular Tekavčić 1975, Larson 1988, 2000, as well as Politzer & Politzer 1953, Sabatini 1965a, b.
8. See Faraoni (forthcoming) for an analysis of all Tuscan documents of the CDL, except those
which the author considered not to be authentic.
9
10
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
c
per futura tempora
“for the times to come”
(CDL I, 170, line 4)
Nevertheless, beside this agreement pattern one finds, in the same texts, 33 occurrences of the innovation type ill-e brachi-a “th(os)e-f.pl arm(n)-pl”:9
(10) a
b.
c.
ad prenominatas baptisteria10
(CDL I, 49–50, lines 12 and 1)
“to the baptisteries already mentioned”
ipse predicte monasteria
(CDL II, 153, line 5)
“the (same) monasteries already mentioned”
ad ipse sanctorum loca
(CDL II, 153, line 7)
“to the places of the saints”
Of course, one might want to interpret the -a agreement morpheme occurring
in (9) as an instance of mere preservation of the Classical Latin system, rather
than data reflecting the Romance variety spoken (and not yet written) at the time.
However, an alternative interpretation seems preferable. The texts collected in the
CDL, at least in the so-called ‘free parts’ of the documents (viz. those that the
notary public wrote without being supported by any fixed repertoire of readymade formulae), have unanimously been judged by the scholars who investigated
them (first and foremost Sabatini 1965a, b)11 as largely mirroring the real spoken
usage of the time. In particular, those texts often presented, as in our case, the
coexistence/competition of more conservative and more innovatory variants, as
predicted in a transitional stage such as that. In the CDL texts examined, the type
illa brachia occurs more frequently than the innovation ille brachia: the ratio is
68:33 occurrences in the texts from which the data in (9)–(10) are drawn. The
two alternate freely in one and the same text within a few lines, as, for example, in
9. See also Larson (1988: §25) and Tekavčić (1975: 227). Earlier examples are found in the Latin
translations of Oribasius (about 600) (see Väänänen 1967: 111) and in inscriptions, e.g. ossa
exterae “bones that do not belong here” (on a Dalmatic inscription; see Tekavčić 1975: 227,
Alkire & Rosen 2010: 195).
10. The form prenominatas counts as an instance of the innovatory type for two reasons: first,
because independently of the case value, the formative -as realizes a feminine gender value (as
a matter of fact, in the CDL, -as and -(a)e may alternate for feminine plural regardless of their
syntactic function), and second, because the Italian formative -e realizing the plural of feminine
nouns belonging to the class casa / case may result from the phonological evolution of -as (e.g.,
Maiden 1996).
11. As Sabatini (1965b: 26) puts it, these legal texts “rappresentano il tipo di scrittura in complesso più sensibile ai fenomeni, fonomorfologici e morfosintattici, dell’uso vivo” (“represent on the
whole the kind of writing more sensitive to phonological, morphological and morphosyntactic
phenomena of living [vulgar] usage”).
The third gender of Old Italian
document no. 194 of the CDL (2nd vol., 183–187), where one finds ipse s(an)c(t)e
loca “the:f.pl holy:f.pl places(n)” (185, line 21) alongside euangelica praeceptas
“Gospel precepts” (184, line 5), where the adjective has the regular n.pl ending,
whereas the noun, which is also neuter and attested in the same text in its classical
form (uiuere secundum D(e)i preceptum “to live according to God’s precept”, 186,
r. 2) takes a hypercorrect ending -s. In sum, the Latin grammatical tradition (prescribing illa brachia) may well have contributed to this conservative type still prevailing quantitatively over the innovation ille brachia. However, all in all, evidence
from several sources backs up the witness of these texts and suggests that the data
in (9)–(10) do mirror genuine linguistic optionality in the gender agreement selected by (those that were to become) nouns of the alternating gender. Beside the
scanty Old Tuscan leftovers in (7), this evidence also comes from other Romance
data, both within and outside of Italo-Romance.
5. The comparative Romance picture
The fact that the evidence for dedicated neuter plural target agreement in Old
Tuscan (see (7)) has escaped all previous studies on Old Italian noun inflection
(see e.g. Maiden 1995, D’Achille & Thornton 2003, Penello et al. 2010: 1389–1397)
may at first sight appear as something of a paradox, given that Old Tuscan is the
most extensively documented and best investigated among the medieval ItaloRomance varieties. Moreover, the existence (or, rather, persistence) of neuter
plural agreement of the kind illustrated in (7) for Old Tuscan had been previously described for other Italo-Romance varieties which — as recalled in fn. 1
above — are independent daughter languages to Latin and hence sister languages
to Old Tuscan (also called, retrospectively, Old Italian). Certainly, the fact that
the Tuscan data in (7) had gone unnoticed has to do with their rarity. On the
contrary, in medieval Italo-Romance varieties from the Center-South, the same
agreement pattern found in Old Tuscan was alive and well throughout the Middle
Ages and is widely attested. This has been shown, in particular for Old Neapolitan,
in Formentin (1998: 291–293) (see also Ledgeway 2009: 149), from which the following examples are drawn:
(11) a. inperò cerasa da epsa Cerer(e) sono chiamata
“therefore, they are called cerasa [“cherries”] from the name of Cerere
herself ”
(Libro de li antichi facti de li gentili o de li pagani, early 15th century)
b. a le dicta mura
“towards the walls mentioned”
11
12
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
(Cronaca di Partenope, ms. I D 14 Bibl. Centr. Regione Siciliana, late
14th century)
The examples in (11) show that the dedicated neuter plural agreement formatives
survived up to the 15th century (and beyond, as shown by the further evidence
gathered in the quoted studies of Old Neapolitan). During the 13th–14th centuries, this agreement pattern was even more robust, as documented by examples
from earlier Neapolitan texts such as Bagni di Pozzuoli in (12) (around 1300, ed.
Pelaez 1928) and Libro de la destructione de Troya in (13) (late 14th century, ed.
De Blasi 1986):12
(12) a.
b.
c.
d.
e.
sola chesta locora ne poteno sanare (15)
“only (adj.) these places can cure us”
chesta predicta omnia (64)
“all these things said (until now)”
chesta bagnora (103)
“these baths”
li homine trovano sua disia (200)
“people find (satisfaction to) their wishes”
doglla face a la latora (243)
“it causes pain in the sides”.
(13) a. quella mura da la parte de fore frabicate e coperta de marmore ben
laborate (78, lines 33–34)
“those outer walls made and covered by well-polished marble”
b. sopervennero la trona spotestata e fuorte (121, line 10)
“there came heavy and strong thunders”
c. a molti erano braza taglyata (173, line 4)
“many of them had got their arms cut”
d. e la mura de quella camera erano facta … et embestuta de deverse petre
preciose (192, lines 13–14)
“and the walls of that room were made … and covered by several
precious stones”
e. per l’ossa de la braza … a la ’strementate de la deta (200, lines 13–14)
“through the bones of the arms … to the tips of the fingers”
f. co la puyna se batteva la face (312, lines 6–7)
“he hit his face with his fists”.
Note that 3pl verb agreement (e.g. poteno in (12a), sopervennero in (13b), erano
in (13c–d)) guarantees that the a-ending noun forms have not been reanalyzed as
12. See Formentin (1998: 292 n. 844, 304, 315–319).
The third gender of Old Italian
feminine singular — a reanalysis that indeed occurred widely across Romance (as
mentioned in §4.2 above).
Presumably, the remaining Italo-Romance varieties of the Center-South, which
are less well documented than Neapolitan, displayed a comparable situation, as the
examples in (14a), from Old Abruzzese, and in (14b), from Old Lucanian, suggest:
(14) a. Ché le nostra molina se non poteano guardare
“Because our mills could not be protected”
(Buccio di Ranallo, Cronaca, about 1362, Old Aquilano, Abruzzese, De
Bartholomaeis 1907: 260)
b. piglia la cotognia … et mondale e bene e piglia mela che non siano bene
fatte, siano uno poco agresta
“take the quinces…and clean them well and take apples which are not
fully ripe [but] a bit unripe”
(Ricettario lucano, 16th century; Süthold 1994: 15, lines 244f)
As is apparent from the data in (14) — and the same could be shown for Old
Neapolitan as well — plural agreement with neuter nouns could vary, as also
feminine plural agreement targets (e.g., (monda)-le in (14a), fatte in (14b)) could
occur alongside those with the dedicated neuter plural a-ending (e.g., nostra in
(14a), la, agresta in (14b)). This optionality appears to have been carried over
from the earlier stage, which can be reconstructed based on the variation in Late
Latin exemplified in (9)–(10). This variation was then resolved at different rates in
different parts of the Romance-speaking territory. In 13th–14th-century Tuscan,
the simplification of the gender distinctions — leading first to the reduction of the
neuter to a controller gender and eventually to its loss — seems to have been more
advanced than in central-southern varieties, as evidenced by the less numerous examples of the neuter plural agreement (see (7) above). In northern ItaloRomance, on the other hand, the reduction was accomplished even earlier, since,
even in the earliest documents, no trace is left of the dedicated neuter agreement
pattern found in the central and southern texts.13 However, this is an argumentum
e silentio built on the record available so far, which need not perforce imply that
this agreement pattern did not exist at all in (old stages of) those varieties. As a
matter of fact, in neighboring Gallo- and Rhaeto-Romance, some few examples
are found.
13. Latin charts from northern Italy, contemporary to those from Tuscany cited in (9)–(10),
show that in a stage previous to the earliest extant Vulgar attestations, the conservative neuter
plural agreement (via the formative -a) persisted in northern Italy as well, though in a smaller
proportion than in Tuscany and central-southern Italy (cf. Faraoni forthcoming).
13
14
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
The Romansh evidence is particularly interesting in this respect, since both
Surselvan and Engadinian preserved dedicated neuter plural agreement until
much later than the other Romance varieties touched upon so far. This led Ascoli
(1880–1883: 439), in his seminal study of Romansh morphosyntax, to claim that
this “è una condizione che non si rinviene se non ne’ Grigioni” (“is a situation
which is found solely in Graubünden [where Romansh is spoken]”), a conclusion
which must nowadays be revised, since the Old Romansh data perfectly parallel
those from Old Tuscan analyzed in §4.2, as well as those from Old Neapolitan
discussed earlier in this section.
Contemporary Romansh has a class of feminine singularia tantum (e.g.
Surselvan la bratscha “the arms”, la feglia “the foliage”, la pera “the pears”, cf.
Lausberg 1976: 25), which derive etymologically from neuter plurals and retain
collective semantics. A remnant of the former status as neuter plurals of these
noun forms persists in their selection of dedicated agreeing forms of the numerals “two” and “three”, distinct from both the masculine and the feminine, as described in modern grammars of several Romansch varieties (see Ganzoni 1977: 56
on Upper Engadinian (15a) and Candinas 1982: 110–111, Spescha 1989: 312–313
on Surselvan (15b)):
(15) a. dua/traia pêra/daunta “two pears/teeth” (vs. dus “two:m=f”) Upper
Engadinian
b. dua/trei pèra “two/three pears” (vs. dus “two:m”, duas “two:f”) Surselvan
In synchronic terms, this kind of marking on lower numerals in modern Romansh,
which departs from the binary gender contrast generally marked on targets elsewhere in the language, is to be analyzed as an instance of gender overdifferentiation of the kind discussed by Corbett (1991: 168), who reports a similar case
of Dravidian languages (Kolami, Ollari, Parji and Naiki), where lower numerals
(“two”, “three” and “four”) have dedicated agreement forms for female human
nouns, in addition to those for male human vs. other, generally found on agreement targets.
However, the abovementioned Romansh a-ending collective/mass nouns
could still select dedicated neuter plural determiner forms (e.g., la “the”, questa
“these”, quella “those”, etc.) and 3pl verb agreement as late as the 17th century, as
exemplified for Engadinian in (16) and Surselvan in (17) (discussed by Velleman
1915: 115–116 and Ascoli 1880–1883: 439, respectively):14
14. Compare also Wilkinson (1985–1991: 41–42), who mentions both the Old Romansh data
with plural agreement (like (16)–(17)) and the modern examples with dedicated forms of the
numerals (15). Concerning the latter, however, he takes issue with Lausberg (1976: 166), who regards those data — correctly, in our view — as “residui arcaici di antiche consecuzioni al neutro
(16) a.
b.
c.
The third gender of Old Italian
La vestimainta sun … cuvertas da la trideza
(Martinus & Rauch 1693: III, 114)
“garments/clothes are … covers of the ugliness”
Ma la mia verva nu vignen à passer via (Bifrun 1560: 171, Marc. 13, 31)
“but my words will not pass away”
E tuotta la nembra nun haun üna proepia houra
(Bifrun 1560: 540, Romans 12, 4)
“and not all the members have the same function”
(17) a. Sia detta han pigliau ilg fijs (Alig 1674: 419)
“Her fingers have taken the spindle”
b. Gual da quei temps vanginen nou navont detta d’ün maun da carstiaun,
ca scriveven … (Bibla 1718: II, 157, Daniel 5.5)
“Just in that moment the fingers of a human hand appeared, which
wrote …”
c. Salidada seias vus, soingia schanuglia (Alig 1674: 262)
“All hail to you, holy knees”
d. … vegnen salvada si la Ss. ossa de S. Placi e S. Sigisbert
(Cuorta Memoria, in Decurtins 1880–1883: 215, lines 20–21)
“… the holy bones of Saint Placidus and Saint Sigebert are bewared”
For instance, in (16a), the NP (la vestimainta) triggers plural agreement on the
verb (sun, rather than es):15 this shows that the determiner la is (neuter) plural
in (16a), and the same can be repeated for the remaining verb forms agreeing
plurale” (“archaic residues of old neuter plural constructions”). Lausberg’s view is the diachronic
counterpart to a synchronic analysis in terms of overdifferentiation. Our fieldwork shows that
while nouns like la pèra “the pears” (as singular feminine collectives) are in themselves stable,
Romansh speakers nowadays find it increasingly difficult to use them in connection with numerals. This confirms the analysis of forms such as dua bratscha as fossilized old neuter forms
(or, synchronically, overdifferentiated gender agreement forms), on the one hand, and discards
Wilkinson’s (1985–1991: 42) analysis in terms of “a single collective unit consisting of two ells”,
on the other. Wilkinson compares Romansh collectives with German “collective nouns beginning with Ge-” such as Gebein “bones” (neuter singular), but fails to see that, in German, zwei
Gebeine does not at all mean “a single collective unit consisting of two bones”, but rather “two
(distinct) sets of bones”, as shown in the following example:
(i) 1671 bekommt der Pfarrer von Schübelbach aus dem Schatz von Einsiedeln zwei Gebeine
der Heiligen Cölestin und Lucidus (“Wie Gebeine aus den Katakomben Roms an den Obersee
kamen”, Zürcher Unterländer, 25.1.2012, www.zuonline.ch/artikel_32770.html) “In 1671, the
priest of the parish Schübelbach receives from the Einsiedeln treasure two [sets of] bones of
Saints Celestine and Lucidus”.
15. The innovative construction with agreement in the (feminine) singular could occur already,
at that time, alongside the conservative neuter plural one.
15
16
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
with subject NPs in (16)–(17), which all show 3pl agreement morphology: vignen,
haun (16b–c), han, vanginen, scriveven (17a–b), vegnen (17d).16
Even this brief review of the comparative Romance evidence makes clear that
the Old Tuscan data discussed in §4 belong to a coherent picture: in particular, it
seems fair to assume that the dedicated neuter agreement formative -a, documented scantily for Tuscan as late as the 13th–14th century, must have occurred earlier
systematically, with nouns belonging to the inflectional classes dito / dita, prato /
pratora, etc. (the greyed-out classes in (5)). Later, it came to be rivaled by the formative -e, which originally realized agreement with plural feminine nouns: in other
words, at this transitional stage, the (partially) conservative three-way target gender
system was in competition with the innovative three-way controller gender system.
By the 13th–14th century, the latter formative prevailed, and the type le braccia dominates over the type la braccia. Thus, the few remnants that we have discovered and discussed in (7) are the last reflexes of the autonomous expression of
the Latin neuter, which persisted into Romance in the intermediate stage (18b), a
stage of preliterary Italian that has to be added to our reconstruction:
(18) a.
Latin
sg
m -us
I
n -um II
f
-a III
b.
pl
-i
-a
-ae
>
preliterary Italian
c.
Old Italian
sg
pl
sg
m
I
-i
m
I
-o
-o
III -a >
III
f
-a
II
-e
f
-a
II
pl
-i
-e
6. Conclusion
Until now, there was broad consensus that Italian, from its earliest medieval documentation, already had a two-gender system like today. In this paper, we have
shown that this traditional view is ill-founded and that Old Italian (Tuscan), as
documented in the late Middle Ages, still possessed a third gender, a successor of
the Latin neuter. In the 13th–14th-century Florentine system, this third gender
was a controller gender (18c) (as defined in (2) above) similar to the neuter that
has been retained to this day in modern Romanian, since nouns assigned to it select agreement forms which are fully syncretic (with masculine in the singular and
with feminine in the plural). However, as late as the 14th century, some remnants
16. As for (17c), while the verb form seias is homophonous with the 2sg in Surselvan (e.g.
Salidada seias, ti zun bialla Faccia literally “all hail to you, his handsome face” in the same enumeration in Alig 1674: 261), the 2pl pronoun form vus guarantees that soingia schanuglia in
(17c) is (morphosyntactically) plural.
The third gender of Old Italian
remained of its previous status as a target gender (18b), as the nouns belonging to
it could still select a dedicated (i.e., non-syncretic) formative for plural agreement.
This older stage, of which Old Tuscan displays just a scant few relics, is much
better preserved in Old Neapolitan and Old Romansh, as shown in §5, whereas
northern Italo-Romance seems to have completed the shift to a two-gender system
earlier than Tuscan.
While our descriptive result concerns specifically Old Italian, the analysis has
some wider implications. First, the reconstruction of stepwise change from three
to two genders from preliterary to modern Italian has consequences for other
Romance languages: it provides comparative evidence strongly suggesting that, for
modern Romanian as well, a three-gender analysis is to be preferred over alternatives, which have been argued for extensively over the past few decades. Proposals
such as Bateman & Polinsky (2010) or Giurgea (2010) arguing explicitly that modern Romanian is best analyzed in terms of two grammatical genders, or Farkas’s
(1990: 539–545) analysis in terms of a binary [±feminine] contrast within a system
in which neuter nouns are not lexically specified as such, theoretically elaborate as
they are, take neither diachronic nor comparative Romance evidence into account
and indeed are much less easy to reconcile with the diachronic and comparative
scenario sketched here than three-gender analyses of Romanian such as Corbett’s
(1991: 151) (or others mentioned in fn. 3). Under this scenario, the gender system
of modern Romanian appears as a natural evolution of a previous stage, such as
the one we have reconstructed for early Old Italian, and which had already been
known to obtain — as we saw in §5 — for Old Neapolitan and Old Romansh. A
byproduct of our discussion is a clear case for the relevance of diachrony for the
assessment (of the merits) of competing synchronic analyses, an argument whose
actual relevance is not fully appreciated in the kind of theoretical literature in formal syntax/morphology just exemplified.
Another seemingly obvious, but nonetheless important, result is that our analysis attests to the effectiveness of the theoretical tools that made it possible, as laid
out in §§2 and 4.1. Old Italian and its grammatical gender system are topics to
which much scholarly effort had been devoted previously, without arriving at what
now clearly appears to be the ‘right’ result: Old Italian had three genders, a fact that
becomes clear when one scrutinizes the primary data anew and especially with the
distinction in mind between target and controller gender.
17
18
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
List of abbreviations
3
f
m
n
sbj
sg
pl
third person
feminine
masculine
neuter
subject
singular
plural
References
Acquaviva, Paolo. 2008. Lexical plurals: A morphosemantic approach. Oxford: Oxford University
Press.
Aikhenvald, Alexandra Y. 2000. Classifiers: A typology of noun categorization devices. Oxford:
Oxford University Press.
Alig, Balthasar. 1674. Epistolas ad Evangelis sin tuttas domeingias, a firaus; a gijs della quareisma,
cun la Passiun de N. S. I. Christi. Cuera si Cuort: Barbisch.
Alkire, Ti & Carol Rosen. 2010. Romance languages: A historical introduction. Cambridge:
Cambridge University Press. DOI: 10.1017/CBO9780511845192
Ascoli, Graziadio Isaia. 1880–1883. Annotazioni sistematiche al Barlaam e Giosafat soprasilvano. Archivio Glottologico Italiano 7. 406–602.
Baldelli, Ignazio. 1973. La carta pisana di Filadelfia. Conto navale pisano dei primi decenni del
sec. XII. Studi di Filologia Italiana 31. 5–33.
Bateman, Nicoleta & Maria Polinsky. 2010. Romanian as a two-gender language. In Donna
Gerdts, John Moore & Maria Polinsky (eds.), Festschrift for David Perlmutter, 41–77.
Cambridge, MA: MIT Press.
Bibla. 1718. La S. Bibla quei ei: Tut la Soinchia Scartira, ner Tuts ils Cudischs d’ilg Veder a Nief
Testament, Cun ils Cudischs Apocryphs. Messa giu Ent ilg Languaig Rumonsch de la Ligia
Grischa tras Anchins Survients d’ilg Plaid da Deus d’ils Venerands Colloquis Sur-a Sut ilg
Guault. Coira: Andrea Pfeffer.
Bifrun, Iachiam (ed.). 1560. L’g Nuof Testamaint da nos Signer Jesu Christi. Basel/Poschiavo:
Jacobus Kündig, Stefano Catani, Dolfino Landolfi. (Reprinted 2001, Bern: Schweizerische
Landesbibliothek).
Bonfante, Giuliano. 1961. Esiste il neutro in italiano? Quaderni dell’Istituto di Glottologia di
Bologna 6. 103–109. (Reprinted in Studii romeni, by Bonfante Giuliano, 1973, 161–170.
Rome: Società accademica romena.).
Bonfante, Giuliano. 1964. Il neutro italiano, romeno e albanese. Acta Philologica della Societas
Academica Dacoromana 3. 27–37. (Reprinted in Studii romeni, by Bonfante Giuliano, 1973,
173–187. Rome: Società accademica romena.).
Bonfante, Giuliano. 1977. Ancora il neutro italiano e romeno. L’Italia Dialettale 40. 287–292.
Candinas, Theo. 1982. Romontsch sursilvan. Grammatica elementara per emprender igl idiom
sursilvan. Cuera: Ligia Romontscha.
The third gender of Old Italian
Castellani, Arrigo. 1958. Frammenti d’un libro di banchieri fiorentini del 1211. Studi di Filologia
Italiana 16. 19–95. (Reprinted in Saggi di linguistica e filologia italiana e romanza (1946–
1976), vol. II by Arrigo Castellani, 1980, 73–140. Rome: Salerno Editrice.)
Castellani, Arrigo. 1976. I più antichi testi italiani. Edizione e commento, 2nd edn. Bologna:
Pàtron.
CDL= Luigi Schiaparelli (ed.). 1929–1933. Codice Diplomatico Longobardo, 2 vols. Rome:
Istituto Storico Italiano.
Corbett, Greville G. 1991. Gender. Cambridge: Cambridge University Press. DOI: 10.1017/
CBO9781139166119
Corbett, Greville G. 2006. Agreement. Cambridge: Cambridge University Press.
Corbett, Greville G. 2008. Determining morphosyntactic feature values: The case of case. In
Greville G. Corbett & Michael Noonan (eds.), Case and grammatical relations: Studies in
honor of Bernard Comrie, 1–34. Amsterdam: John Benjamins. DOI: 10.1075/tsl.81.01det
Corbett, Greville G. 2011. The penumbra of morphosyntactic feature systems. In Jonathan
Bobaljik, Uli Sauerland & Andrew Nevins (eds.), Markedness and underspecification in the
morphology and semantics of agreement. Morphology 21(2). 445–480.
D’Achille, Paolo & Anna M. Thornton. 2003. La flessione del nome dall’italiano antico all’italiano
contemporaneo. In Nicoletta Maraschio & Teresa Poggi Salani (eds.), Italia linguistica anno
Mille, Italia linguistica anno Duemila. Atti del XXXIV congresso internazionale di studi della
Società di Linguistica Italiana, Firenze, 19–21 ottobre 2000, 211–230. Rome: Bulzoni.
De Bartholomaeis, Vincenzo (ed.). 1907. Buccio di Ranallo, Cronaca aquilana rimata di Buccio
di Ranallo di Popplito di Aquila. Rome: Istituto Storico Italiano.
De Blasi, Nicola (ed.). 1986. “Libro de la destructione de Troya”. Volgarizzamento napoletano
trecentesco da Guido delle Colonne. Rome: Bonacci.
Decurtins, Caspar. 1880–1883. Quattro testi soprasilvani/Vier sürselvische Texte. Archivio
Glottologico Italiano 7. 149–364.
Della Valle, Valeria. 1982. Le lettere volgari di Filippo Belforti vescovo di Volterra (1348–1353).
Studi Linguistici Italiani 8. 155–263.
Dressler, Wolfgang U. 2003. Degrees of grammatical productivity in inflectional morphology.
Italian Journal of Linguistics 15(1). 31–62.
Dressler, Wolfgang U. & Lavinia Merlini Barbaresi. 1994. Morphopragmatics: Diminutives
and intensifiers in Italian, German and other languages. Berlin: Mouton de Gruyter. DOI:
10.1515/9783110877052
Dressler, Wolfgang U. & Anna M. Thornton. 1996. Italian nominal inflection. Wiener
Linguistische Gazette 55–57. 1–26.
Faraoni, Vincenzo. Forthcoming. Manifestazioni del neutro italo-romanzo nella documentazione notarile altomedievale. In Éva Buchi, Jean-Paul Chauveau & Jean-Marie Pierrel (eds.),
Actes du XXVIIe Congrès international de linguistique et de philologie romanes (Nancy, 15–
20 juillet 2013), 3 vols. Strasbourg: Société de linguistique romane/ÉliPhi.
Faraoni, Vincenzo, Francesco Gardani & Michele Loporcaro. 2013. Manifestazioni del neutro
nell’italo-romanzo medievale. In Emili Casanova Herrero & Cesáreo Calvo Rigual (eds.),
Actas del XXVI Congreso Internacional de Lingüística y Filología Románicas (Valencia, 6–11
septiembre 2010), vol. 2, 171–182. Berlin: De Gruyter Mouton.
Farkas, Donka F. 1990. Two cases of underspecification in morphology. Linguistic Inquiry 21.
539–550.
Formentin, Vittorio (ed.). 1998. Loise de Rosa, Ricordi. Rome: Salerno Editrice.
19
20 Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
Ganzoni, Gian Paul. 1977. Grammatica ladina. Grammatica sistematica dal rumantsch d’Engiadin’
Ota per scolars e creschieus da lingua rumauntscha e tudascha. Samedan, Switzerland: Lia
Rumantscha.
Gardani, Francesco. 2013. Dynamics of morphological productivity: The evolution of noun classes
from Latin to Italian. Leiden: Brill. DOI: 10.1163/9789004244658
Giurgea, Ion. 2010. Romanian genderless pronouns and parasitic gaps. In Sonia Colina, Antxon
Olarrea & Ana Maria Carvalho (eds.), Romance linguistics 2009: Selected papers from the
39th LSRL, 231–248. Amsterdam: John Benjamins.
Graur, Alexander. 1928. Les substantifs neutres en roumain. Romania 54. 249–260.
Hall, Robert. 1965. The “neuter” in Romance: A pseudo-problem. Word 21. 421–427.
Hockett, Charles F. 1958. A course in modern linguistics. New York: Macmillan.
Igartua, Iván. 2006. Genus alternans in Indo-European. Indogermanische Forschungen 111. 56–
70.
Jakobson, Roman. 1971. On the Rumanian neuter. In Selected writings, vol. II, World and language, 187–189. The Hague: Mouton.
Kuryłowicz, Jerzy. 1964. The inflectional categories of Indo-European. Heidelberg: Winter.
Larson, Pär. 1988. Gli elementi volgari nelle carte del “Codice Diplomatico Longobardo”. Florence:
University of Florence MA thesis.
Larson, Pär. 2000. Tra linguistica e fonti diplomatiche: quello che le carte dicono e non dicono. In Jószef Herman & Anna Marinetti (eds.), La preistoria dell’italiano. Atti della Tavola
Rotonda di Linguistica Storica (Università Ca’ Foscari di Venezia, 11–13 giugno 1998), 151–
166. Tübingen: Niemeyer.
Lausberg, Heinrich. 1976. Linguistica romanza, 2 vols., 2nd edn. Milan: Feltrinelli.
Ledgeway, Adam. 2009. Grammatica diacronica del napoletano. Tübingen: Niemeyer. DOI:
10.1515/9783484971288
Levasti, Arrigo (ed.). 1924–1926. Beato Iacopo da Varagine, Leggenda Aurea. Volgarizzamento
toscano del Trecento. Florence: Libreria Editrice Fiorentina.
Lisini, Alessandro (ed.). 1903. Il Costituto del comune di Siena volgarizzato nel MCCCIXMCCCX. Siena: Tip. e Lit. Sordomuti di L. Lazzeri.
Loporcaro, Michele & Tania Paciaroni. 2011. Four gender-systems in Indo-European. Folia
Linguistica 45(2). 389–434. DOI: 10.1515/flin.2011.015
Maiden, Martin. 1995. A linguistic history of Italian. London: Longman.
Maiden, Martin. 1996. On the Romance inflectional endings -i and -e: A methodological issue.
Romance Philology 50(2). 147–182.
Matasović, Ranko. 2004. Gender in Indo-European. Heidelberg: Winter.
Martinus, Johannes & Andreas V. Rauch. 1693. Abyss dal’aeternitat in 3 cudeschs considerata.
Turi [Zurich]: Simler.
Menichetti, Aldo (ed.). 1965. Chiaro Davanzati, Rime. Bologna: Commissione per i testi di lingua.
Merlo, Clemente. 1952. Ital. le labbra, le braccia e sim. Italica (Bulletin of the American Society of
Teachers of Italian) 29(4). 229–234.
Negroni, Carlo (ed.). 1886. La Bibbia volgare secondo la rara edizione del I di ottobre MCCCCLXXI,
vol. 9: I Quattro Evangeli e gli Atti degli Apostoli, i Maccabei. Bologna: Romagnoli.
Pelaez, Mario. 1928. Un nuovo testo dei Bagni di Pozzuoli in volgare napoletano. Studj Romanzi
19. 47–134.
The third gender of Old Italian
Penello, Nicoletta, Paola Benincà, Laura Vanelli & Roberta Maschi. 2010. Morfologia flessiva.
In Giampaolo Salvi & Lorenzo Renzi (eds.), Grammatica dell’italiano antico, 1389–1491.
Bologna: Il Mulino.
Pernicone, Vincenzo (ed.). 1946. Franco Sacchetti, Il Trecentonovelle. Florence: Sansoni.
Polidori, Filippo-Luigi (ed.). 1845. Rosaio della vita, trattato morale attribuito a Matteo de’
Corsini e composto nel MCCCLXXIII. Florence: Società Poligrafica Italiana.
Politzer, Frida & Robert L. Politzer. 1953. Romance trends in 7th and 8th century Latin documents. Chapel Hill: The University of North Carolina Press.
Rovai, Francesco. 2012. Between feminine singular and neuter plural: Reanalysis patterns. Transactions of the Philological Society 110(1). 94–121. DOI: 10.1111/j.1467–
968X.2012.01286.x
Sabatini, Francesco. 1965a. Esigenze di realismo e dislocazione morfologica in testi preromanzi. Rivista di Cultura Classica e Medioevale 7 (Studi in onore di A. Schiaffini), 972–988.
(Reprinted in Italia linguistica delle origini. Saggi editi dal 1956 al 1996 ed. by Vittorio Coletti,
Rosario Coluccia, Paolo D’Achille & Nicola De Blasi, 1996, 99–131. Lecce, Italy: Argo.)
Sabatini, Francesco. 1965b. Sull’origine dei plurali italiani: il tipo in -i. Studi Linguistici Italiani 5.
5–39. (Reprinted in Italia linguistica delle origini. Saggi editi dal 1956 al 1996 ed. by Vittorio
Coletti, Rosario Coluccia, Paolo D’Achille & Nicola De Blasi, 1996, 133–172. Lecce, Italy:
Argo.)
Salvi, Giampaolo & Lorenzo Renzi (eds.). 2010. Grammatica dell’italiano antico. Bologna: Il
Mulino.
Schön, Ilse. 1971. Neutrum und Kollektivum. Das Morphem -a im Lateinischen und Romanischen.
Innsbruck: Institut für Vergleichende Sprachwissenschaft der Universität Innsbruck.
Spescha, Arnold. 1989. Grammatica sursilvana. Cuera, Switzerland: Casa editura per mieds
d’instrucziun.
Süthold, Michael. 1994. Manoscritto Lucano: Ein unveröffentlichtes Kochbuch aus Süditalien vom
Beginn des 16. Jahrhunderts. Geneva: Droz.
Tekavčić, Pavao. 1975. Agli albori dell’italiano. Linguistica 15. 209–239.
Tekavčić, Pavao. 1980. Grammatica storica dell’italiano, 2 vols. Bologna: Il Mulino.
Togeby, Knud. 1952. Le problème du neutre roumain. Cahiers Sextil Puşcariu 1(2). 265–268.
Väänänen, Veikko. 1967. Introduction au latin vulgaire, 2nd edn. Paris: Klincksieck.
Velleman, Antonius. 1915. Grammatica teoretica, pratica ed istorica della lingua ladina
d’Engiadin’ Ota, vol. 1: Il Substantiv, l’Artichel, l’Aggetiv, il Pronom. Zurich: Füssli.
Wilkinson, Hugh. 1985–1991. The Latin neuter plurals in Romance (I–VII). Ronshu 26. 137–
150, 27.157–171, 28.33–46, 29.47–61, 30.109–122, 31.113–127, 32.35–50, collected in
http://www6.ocn.ne.jp/~wil/lnp.pdf (12 August 2012).
Résumé
On démontre ici que l’ancien italien avait un système à trois genres, dans lequel le neutre était
encore une valeur à plein titre de cette catégorie. Les preuves que nous apportons montrent (a)
qu’en ancien italien il y avait trois séries de noms contrôleurs, qui sélectionnaient trois schémas d’accord distincts ; (b) qu’à chacun de ces ensembles de contrôleurs, y compris le neutre,
des noms étaient assignés qui appartenaient à des classes flexionnelles différentes, toutes productives ; et enfin (c) que le neutre sélectionnait encore au moins un formatif dédié marquant
21
22
Michele Loporcaro, Vincenzo Faraoni and Francesco Gardani
l’accord, qui représentait le reste de son ancien statut de ‘genre cible’. Ces preuves nouvelles tirées
de l’ancien italien cadrent bien avec ce qu’on sait des stades révolus d’autres variétés romanes.
Enfin, nous nous penchons brièvement sur les conséquences de nos résultats pour, d’une part,
la reconstruction de la transition latino-romane, ainsi que (de façon plus générale) pour l’approche théorique et méthodologique de l’étude du développement diachronique des systèmes
de genre.
Zusammenfassung
Dieser Aufsatz zeigt, dass das Altitalienische ein Drei-Genus-System hatte, in dem das
Neutrum noch ein vollqualifizierter Genuswert war. Die diese Annahme begründenden
Daten zeigen (a), dass es im Altitalienischen drei unterschiedliche Sets von Controllers gab,
von denen jedes ein eigenes Kongruenzmuster selegierte; (b), dass jedem dieser drei Sets von
Controllers, einschließlich des Neutrums, Nomina zugewiesen waren, die unterschiedlichen
produktiven Flexionsklassen angehörten; und (c), dass das Neutrum nach wie vor spezielle
Kongruenzmorpheme selegierte — die Reste seines ursprünglichen Status als ‘target gender’.
Diese neuartigen Daten über das Altitalienische stimmen mit Belegen von älteren Sprachstufen
anderer romanischen Varietäten überein. Der Aufsatz bespricht die Folgen dieser Erkenntnisse
sowohl für die Rekonstruktion der Übergangsphase zwischen Latein und Romanisch, als
auch für den theoretischen und methodologischen Ansatz zur Untersuchung der diachronen
Entwicklung von Genussystemen allgemein.
Authors’ addresses
Michele Loporcaro
Universität Zürich
Romanisches Seminar
Zürichbergstrasse 8
8032 Zürich
Switzerland
Vincenzo Faraoni
Universität Zürich
Romanisches Seminar
Zürichbergstrasse 8
8032 Zürich
Switzerland
loporcar@rom.uzh.ch
vfaraoni@rom.uzh.ch
Francesco Gardani
Wirtschaftsuniversität Wien
Institut für Romanische Sprachen
Welthandelsplatz 1
1020 Wien
Austria
francesco.gardani@wu.ac.at